PureBasic Forums - English

the.weavster wrote:

purenet wrote:Python app is 60-300 times slower than PB app and uses 2-3 times more memory

Could you provide some examples demonstrating that?

Not direct comparison, but I converted some Java programs to PB and got significant performance gain (2-3 times faster).
Comparison is possible, because I didn't change program logic, I had no time to hunting new bugs resulted of big changes.
Those programs process high amount of data (gigabytes) in memory. Happy surprise was that memory consumption dropped to 20% of earlier.
This own experience is in line what scientists reported below, there is also Python on the list.
PB is not far from C performance but generally slightly slower, less compiler optimizations.

https://benchmarksgame-team.pages.debia ... 3-gcc.html
https://www.researchgate.net/publicatio ... ory_relate

This is going a bit off-topic, anyway exciting to see how fast PB for ARM will be.

purenet wrote:

the.weavster wrote:

purenet wrote:Python app is 60-300 times slower than PB app and uses 2-3 times more memory

Could you provide some examples demonstrating that?

Not direct comparison, but I converted some Java programs to PB and got significant performance gain (2-3 times faster).
Comparison is possible, because I didn't change program logic, I had no time to hunting new bugs resulted of big changes.
Those programs process high amount of data (gigabytes) in memory. Happy surprise was that memory consumption dropped to 20% of earlier.
This own experience is in line what scientists reported below, there is also Python on the list.
PB is not far from C performance but generally slightly slower, less compiler optimizations.

https://benchmarksgame-team.pages.debia ... 3-gcc.html
https://www.researchgate.net/publicatio ... ory_relate

This is going a bit off-topic, anyway exciting to see how fast PB for ARM will be.

Is there a plan when the first version of PB can be tested for the Pi?

derschutzhund wrote:Is there a plan when the first version of PB can be tested for the Pi?

I would imagine that it's still quite a way off. Let's be patient and let Fred do his groovy thang.

derschutzhund wrote:Is there a plan when the first version of PB can be tested for the Pi?

Nobody said officially that PB is coming to Raspberry Pi.

On a related sidenote - the Galaxy Book S seems to be a pretty decent ARM based windows 10 Notebook. The test is in german but it basically says that the notebook is decently fast, light and thin. The downside is that it only emulates x86 32bit and not 64bit.

Bitblazer wrote:The downside is that it only emulates x86 32bit and not 64bit.

The other downside is that emulation is via software, not hardware. Although with octacore processors becoming more common, the emulation could easily be offloaded to its own core(s).

That said, I was strongly looking at ARM based Windows systems this past Christmas, when I was buying a new system.

Again, brilliant of Fred and PB will NOT be left in the dust now that Apple has announced they will be moving Macs to their own ARM-based processors. The macOS version of PB will need to be supporting ARM come 2021! Well-done, Fred!!!!

Kuron wrote:Again, brilliant of Fred and PB will NOT be left in the dust now that Apple has announced they will be moving Macs to their own ARM-based processors. The macOS version of PB will need to be supporting ARM come 2021! Well-done, Fred!!!!

My hope is still a support for the Raspberry Pi! Although I have a license for Purebasic of course I would pay extra for it! The Rapberrys are really out in millions and if PB could program them it would be a whole new market for PB!

derschutzhund wrote:

Kuron wrote:Again, brilliant of Fred and PB will NOT be left in the dust now that Apple has announced they will be moving Macs to their own ARM-based processors. The macOS version of PB will need to be supporting ARM come 2021! Well-done, Fred!!!!

My hope is still a support for the Raspberry Pi! Although I have a license for Purebasic of course I would pay extra for it! The Rapberrys are really out in millions and if PB could program them it would be a whole new market for PB!

Me too.
I've been struggling to find a language for the pi that I like.

If Fred delivers this it will be like an extra Xmas.

Thank you, Fred, in anticipation.

Since 2003, ARM is a big illusion. All this conflict just for a patent about the content access memory.

My next smartphone will be an Intel, and I hope AMD will follow....

I consider good however if PureBasic would allow to code in ARM Assembly, as we can do it actually, on X86/X64 Assembly. But Fred will meet lots of useless technical walls.

Normally, the next hardwared breakthrough should be "to talk directly to the processors", (byte code or word code) and use the last DAC/ADC technologies, exactly as gyroscopic devices (which are analogic).

Olli wrote:My next smartphone will be an Intel, ....

Yeaaaa, the list of current Intel smartphones is very, very impressive

ARM has been a lot more power efficient in the past, compared to the x86 architecture. Thats crucial for mobile devices where constant processing power is not as important. You can find a lot discussions about the "performance per watt" comparision on the net.

Olli wrote:My next smartphone will be an Intel, and I hope AMD will follow....

https://www.anandtech.com/show/12592/in ... -benchmark

Fred wrote:I shouldn't say it because it's way to early (no ETA, only very small programs works), but I'm actually working on it ! Now, back to code

Wow that is awesome news!
I will happily send a donation for that. Together with all the good contributions to the IDE I am very eager to see the next purebasic versions.

Fred wrote:I shouldn't say it because it's way to early (no ETA, only very small programs works), but I'm actually working on it ! Now, back to code

ARM != ARM

Here a conversation of "The Pirates of Apple Silicon":

https://www.youtube.com/watch?v=t1pdnQR ... e=youtu.be

These guys did a tremendous job getting their compiler/toolchain to work on Apple Silicon & macOS.

Fred,
The first 35 minutes are the important ones.
Hope they share valuable insight for you...

Take care.

Fred wrote:I shouldn't say it because it's way to early (no ETA, only very small programs works), but I'm actually working on it

Any news?